Nvidia will Make Chips for CPU-GPU Supercomputing

The Nvidia releases the source code for CUDA compiler

The heart of business is the manufacture of Nvidia graphics chip performance. The generalization of graphics chips in Intel Nvidia forced to refocus its activity on two points:

  • The first is the production of graphics optimized for compute-intensive, these products are cutting a great success.
  • Almost opposite the company released its Tegra chips, for mobile devices and with a heart ARM plus a generous graphics.

Nvidia will move to a new merging somehow the two ideas in order to become completely independent from Intel and to a lesser extent AMD in data centers. Indeed, its GP-GPU cards have beautiful develop enormous power, they remain dependent on central processors for the drive, assign tasks …. which confine them to the status of devices or co-processors.

The next step will be announced by NVIDIA to integrate one or more CPU cores in its GP-GPU cards.Thus, they become autonomous. The CPU cores will of course be based on the ARM architecture, the V8 to be more precise, which supports 64-bit.

This announcement should give more credit to the bright future predicted ARM chips in its data centers currently monopolized mostly by Intel.

Spread the love